Paradigmatic opacity in Nuer - uni-leipzig.de · Matthew Baerman, Surrey Morphology Group, University of Surrey [email protected] revised post-conference version of 14/7/09 1

The support of the European Research Council (grant ERC-2008-AdG-230268 MORPHOLOGY) is gratefully

acknowledged.

Paradigmatic opacity in Nuer

Matthew Baerman, Surrey Morphology Group, University of Surrey

[email protected]

revised post-conference version of 14/7/09

1 Paradigm Economy

Is there an upper limit on the number of inflection classes a system can maintain?

(1) Paradigm Economy Principle (paraphrased) (Carstairs 1983): the upper limit does not

exceed (by very much) the logical limit needed to account for the allomorphy of any single

value.

(2) I II III

1 a a a

2 b c c

3 d e f

(3) No Blur Principle (Carstairs-McCarthy 1994): Within any set of competing inflectional

affixal realizations for the same paradigmatic cell, no more than one can fail to identify

inflection class unambiguously.

system with no blur system with massive blur

(4) I II III IV I II III IV

1 a b b b 1 a a b b

2 c d c c 2 c d c d

3 e e f e 3 e f f e

These have proved to be too restrictive (see in particular Finkel & Stump 2007).

(5) Inflection Class Economy Theorem (Müller 2007): Given a set of n inflection markers,

there can be at most 2n−1 inflection classes, independently of the number of instantiations of

the grammatical category that the markers have to distribute over.

(6) The gist of the proposal:

a. markers may be linked to multiple values (Syncretism) b. any competion between markers is resolved by a rule hierarchy (Specificity) c. inflection classes consist of a list of the markers not used by a given lexeme

(7) Markers Specificity Inflection classes Resulting paradigm

a = elsewhere

b = 2

b > a

I has no constraints

II does not use b

I II

1 a a

2 b a

Given two markers, the maximal system has two values realized across two inflection classes. By

contrast, without such constraints, you could generate an endless number of values and inflection

classes just by jumbling around the distribution of (a,b).

a

d

c b

e f

value 1

value 2

value 3

Morphology of the World’s Languages

Leipzig, June 11-13, 2009

2

(8)

2 v

alu

e I II III IV

1 a a b b

2 a b a b

(9)

3 v

alu

e

I II III IV V VI VII VIII

1 a a a a b b b b

2 a a b b a a b b

3 a b a b a b a b

(10)

4 v

alu

e

I II III IV V VI VII VIII IX X XI XII XIII XIV XV XVI

1 a a a a a a a a b b b b b b b b

2 a a a a b b b b a a a a b b b b

3 a a b b a a b b a a b b a a b b

4 a b a b a b a b a b a b a b a b

2 Nouns in Nuer

(11) Genetic affiliation (Storch 2005: 17, based on earlier sources)

Nilo-Saharan

Eastern Sudanic

Nilotic

West Nilotic

Dinka Nuer Atuot

(12) Inflectional devices (data from Frank 1999)

ending only stem + ending stem alternation only

NOM SG jith läm jiath GEN SG jith-kä läm-kä jiaath LOC SG jith-kä lam-kä jiaath

NOM PL jith-ni läm-ni jiεn

GEN PL jith-ni lääm-ni jiɛɛn LOC PL jith-ni lam-ni jiεεn

‘ear’ ‘rank’ ‘kind of tree’

3

(13) Case-number endings in Nuer nouns (Frank 1999). Corpus of 252 complete noun

paradigms. Out of four affixes (Ø, -ä, -kä, -ni )1 we get 24 classes – the ICET predicts 16.

I II III IV V VI VII VIII IX X XI XII

NOM SG Ø Ø Ø Ø Ø Ø Ø Ø Ø Ø Ø Ø

GEN SG Ø kä kä Ø Ø Ø Ø kä kä kä kä kä

LOC SG Ø kä kä Ø Ø Ø kä Ø Ø kä kä kä

NOM PL Ø Ø ni ni Ø Ø Ø ni Ø Ø Ø Ø

GEN PL ni ni ni ni Ø ni ni ni ni Ø ni Ø

LOC PL ni ni ni ni Ø Ø ni ni ni Ø Ø ni

% of corpus: 24 21 18 10 4 4 4 4

XIII XIV XV XVI XVII XVIII XIX XX XXI XXII XXIII XXIV


GEN SG kä kä Ø Ø kä ä ä ä kä kä Ø Ø

LOC SG Ø kä kä kä kä ä ä kä ä ä ä ä

NOM PL Ø ni ni Ø Ø ni Ø Ø ni Ø ni Ø

GEN PL ni Ø ni Ø kä ni ni ni ni ni ni ni

LOC PL Ø ni ni ni ni ni ni ni ni ni ni ni

If we leave aside -kä ~ -ä allomorphy (and the very odd class XVII), then it’s 16 classes against an ICET prediction of 8, i.e. most of the action involves Ø, -kä, -ni .

(14) The 16 classes can be broken down into combinations of 4 singular patterns and 6 plural

patterns.

1 2 3 4 NOM SG Ø Ø Ø Ø GEN SG Ø -kä -kä Ø LOC SG -kä -kä Ø Ø

5 6 7 8 9 10 NOM PL -ni Ø -ni Ø Ø Ø

GEN PL Ø Ø -ni -ni -ni Ø

LOC PL -ni -ni -ni -ni Ø Ø

There is a great variety of stem alternation processes and patterns, but distribution of affixes is largely independent of this:

1 Also: plural -ni regularly shows the allomorph -i after -l, -n and -r.

V

I

IV

XI

XII X

IX VII

III

VIII

XVI

II

XIII

XV

singular

patterns VI

XIV

XV

VI

XIV

plural

patterns



4

(15) Examples of singular paradigms (invariant stems)

1 2 3 4

NOM SG nyaŋyεt liεth mɛẗh pät GEN SG nyaŋyεt liεth-kä mɛẗh pät-kä LOC SG nyaŋyεt-kä liεth mɛẗh pät-kä ‘ring’ ‘fat/oil’ ‘kind of tree’ ‘slap’

(16) Examples of plural paradigms (invariant stems)

5 6 7 8 9 10

NOM PL barkaɣ-ni cak gaak-ni wiiɣ kεεc tεεr GEN PL barkaɣ cak gaak-ni wiiɣ-ni kεεc-ni tεεr LOC PL barkaɣ-ni cak-ni gaak-ni wiiɣ-ni kεεc tεεr ‘hoof’ ‘milk’ ‘flower’ ‘village’ ‘kind of tree’ ‘conflict

Crazzolara (1933) and Vandevort (n.d.) describe what appear to be somewhat different varieties of Nuer, which share the following:

o GEN/LOC SG -kä ~ -ä allomorphy favours -ä (mirror image of Frank 1999 material) o Case endings always syncretize GEN and LOC.

Allowing for these differences, all three varieties show the same major patterns with the same relative frequency. Upshot: we can be confident that at least these are genuine patterns.

(17) Frank (1999); corpus = 252 nouns

I II III IV V X

NOM SG Ø Ø Ø Ø Ø Ø

GEN/LOC SG Ø kä kä Ø Ø kä

NOM PL Ø Ø ni ni Ø Ø

GEN/LOC PL ni ni ni ni Ø Ø

% of corpus: 24 21 18 10 4 ---

(18) Crazzolara (1933); corpus = 195 nouns

I II III IV V X


GEN/LOC SG Ø ä ä Ø Ø ä



% of corpus: 46 24 8 11 7 4

(19) Vandevort (n.d.); corpus = 329 nouns

I II III IV V X


GEN/LOC SG Ø ä ä Ø Ø ä



% of corpus: 39 27 15 11 6 2

5

This yields 6 classes, against a ICET prediction of 8 – but only by accident! The problem in a nutshell:

(20) According to the terms of the ICET, varying distribution of -ni in the NOM PL can only be described by positing accidently homophonous Ø1 and Ø2 or -ni 1 and -ni 2.

I IV

NOM SG Ø Ø

GEN/LOC SG Ø vs. Ø

NOM PL Ø ni

GEN/LOC PL ni ni

(21) Nor can impoverishment help: NOM PL stem may be distinct both from SG and from the rest

of PL, independent of presence or absence of -ni (data from Frank 1999).

NOM SG jiɔm thiik GEN SG jiam thiak LOC SG jiam thiak NOM PL jɔam thik-ni GEN PL jiam-ni thiak-ni LOC PL jiam-ni thiak-ni ‘wind’ ‘door’

3 Conclusion

Constraints on paradigm structure do not seem to work. Perhaps a more realistic way of thinking

about this is through principal parts (e.g. Finkel & Stump 2007; Ackerman, Blevins & Malouf, to

appear). Inflection class constraints would then translate into constraints on the number of forms

you have to memorize (which is unlikely to make anybody happy).

(22) Dynamic principal parts of classes I-XVI (in terms of Finkel & Stump 2007); shaded cells

are principal parts. 2-5 prinicple parts needed.

I II III IV V VI VII VIII IX X* XI XII







XIII XIV XV XVI*

NOM SG Ø Ø Ø Ø

GEN SG kä kä Ø Ø

LOC SG Ø kä kä kä

NOM PL Ø ni ni Ø

GEN PL ni Ø ni Ø

LOC PL Ø ni ni ni

*an alternative set of 3 principal parts is also possible



6

On the reasonable assumption that there are some default rules (or Paradigm Structure Conditions in Wurzel’s sense), this can be simplified quite a bit. But it’s still pretty involved

for a system with just 3 affixes.

(23) Let’s take class II as the default, which we can derive by the following rules:2

i. default ending = Ø ii. GEN SG = -kä iii. GEN PL = -ni iv. GEN LOC (a rule of referral: the locative is the same as the genitive)

(24) Revised dynamic principal parts of classes I-XVI, assuming the rules in (23). Shaded cells

indicate principal parts that override the default. Gets as high as 4 principal parts.

I II III IV V VI VII VIII IX X XI XII







XIII XIV XV XVI

NOM SG Ø Ø Ø Ø

GEN SG kä kä Ø Ø

LOC SG Ø kä kä kä

NOM PL Ø ni ni Ø

GEN PL ni Ø ni Ø

LOC PL Ø ni ni ni

Perhaps this complexity is licensed by the complexity of the stem alternations (see fig. 25), which must be memorized to a great extent (and which no PEC will rein in anyway). If

you’re memorizing stems, you’re presumably memorizing the endings that go with them.

Note that while the stem alternation pattern and the affixal patterns are largely independent

(see figs 15 and 16), there appears to be general rule which allows for some limited

predictions. In Frank’s corpus, for any given case, the singular and plural forms are different

in all but one example. Thus, where there is no stem alternation marking number, it is

marked by an affix. This helps predict the choice of the common classes I vs. IV and II vs.

III for some lexemes. However, the reverse implication does not hold: where there is a stem

alternation marking number, the choice of affixal pattern is unpredicatable.

News flash: an computational principal parts analysis of all this has been implemented. Come to the

LAGB meeting in Edinburgh to hear more.

2 In fact, Frank identifies class III as the default for novel items, which is distinguished by having –ni in the nominative

plural. But note that novel items also lack a stem alternation, and so would be predicted to have –ni in the nominative

plural in any case. (Obviously, a more complete formal analysis would have to incorporate this.) I have chosen class II

to illustrate a default analysis since it allows for a greater reduction in principal parts.

7

Note on orthography (breathy voice vowels)

ä = IPA a ̤a = IPA æ̤ ë = IPA e ̤i = IPA i ̤ɛ ̈= IPA ɛ ̤ɔ = IPA ɔ ̤

References

Ackerman, F., J. Blevins and R. Malouf, to appear. Parts and wholes: Implicative patterns in

inflectional paradigms. In: J. P. Blevins and J. Blevins (eds) Analogy in grammar: form and

acquisition. Oxford: Oxford University Press.

Carstairs, A. 1983. Paradigm economy. Journal of Linguistics 19.115-28.

Carstairs-McCarthy, A. 1994. Inflection classes, gender, and the Principle of Contrast. Language

70/4. 737-788.

Crazzolara, J.P. 1933. Outlines of a Nuer grammar. Vienna: Anthropos.

Finkel, R. and G. Stump. 2007. Principal parts and degrees of paradigmatic transparency. Technical

Report No. TR 470-07, Department of Computer Science, University of Kentucky.

Frank, W. J. 1999. Nuer noun morphology. M.A. thesis, State University of New York, Buffalo.

Müller, G. 2007. Notes on paradigm economy. Morphology 17/1. 1-38.

Storch, A. 2005. The noun morphology of Western Nilotic. Cologne: Köppe.

Vandevort, E. n.d. Nuer field notes. Scanned index cards available at

http://www.dlib.indiana.edu/collections/nuer/.

.

(25) Stem alternation patterns in Nuer nouns (data from Frank 1999): up to 5 stems, at least 47 different patterns attested. Percent of corpus (252 nouns) not calculated for types with 10 tokens or fewer.

A A A A A A A B B A B B A B B A A A A A A A A A A A A A A A A A A A A B

A A A B B B C C C B B B C D D A A B A B B B A A B B C B C B B C C A C B

NOM SG baan ŋiɛc ley kɔm thin böw ɣöth juac rɔany lath käk läm GEN SG baan-kä ŋiɛc-kä läɣ kɔam thiän böw-kä ɣöth-kä juac-kä rɔany-kä lath käk-kä läm-kä LOC SG baan ŋiɛc-kä läɣ kɔam thiän böw-kä ɣöth-kä juac-kä rɔany-kä lath käk-kä lam-kä NOM PL baa-ni ŋiic leey kɔam-ni thin böw-ni ɣöth-ni juaac rɔny laath-ni kääk läm-ni GEN PL baa-ni ŋiic-ni leey-ni kɔam-ni thii-ni böw-ni ɣööth-ni juac-ni rɔny-ni laath-ni kaak-ni lääm-ni LOC PL baa-ni ŋiic-ni leey-ni kɔam-ni thii-ni bööw-ni ɣööth-ni juac-ni rɔɔny laath-ni kaak-ni lam-ni ‘bug’ ‘ant’ ‘animal’ ‘chair’ ‘breast’ ‘goat’ ‘basket’ ‘grass’ ‘cane’ ‘cotton’ ‘cough’ ‘rank’ 27% 20% 15% 6% 4%

A A B A A B A A B A A B A A B A A B A A B A A B A B A A B A A B A B A A B A B B B B B B C C C C C C D C D C C D D A A A A A C A C D

NOM SG lil deth cuar gɔɔk puäth wäär jiom nyal ŋulciök kɔar took GEN SG lil-kä deth cuar-kä gɔɔk-kä puäth-kä wäär-kä jiom nyal ŋuulciök kɔr-kä tɔktɔk-kä LOC SG liɛl deeth cueer gɔak puth-kä waar jiɔm nyaal ŋulciö-kä kɔar took-ä NOM PL lil-i deeth cueer gɔak-ni puɔth wer jiam nyiär ŋulciök-ni kɔar-i took-ni GEN PL lil-i deth-ni cueer-i gɔak-ni puɔth-ni wer-i jiɔm-ni nyiät ŋulciök-ni kɔar-i tɔktɔk-ni LOC PL lil-i deeth-ni cueer-i gɔaak-ni puɔth-ni wär jiam nyiät ŋulciök-ni kɔaar-i tɔktɔɔk-ni ‘oxbow lake’ ‘load’ ‘sky’ ‘monkey’ ‘lung’ ‘dung’ ‘Nov-Jan’ ‘girl’ ‘heel’ tree’ ‘pail’

A B A A B A A B A A B B A B B A B B A B B A B B A B B A B B A B B A B B

B B B C A A C C B A A B A B B B A A B B C B C B B C C C A A C A C C B B

NOM SG kɔat riöp luak cäŋ libɛ mät pan täp tiik dɛy puɔk jiɔm GEN SG kɔt-kä röp luaak cääŋ lipɛ-kä mat-kä paan tap-kä tiɛk̈ dɔa puk-kä jiam LOC SG kɔata riöp-kä luak cääŋ lipɛ-kä mat-kä paan tap-kä tiɛk̈ dɔa puk-kä jiam NOM PL kɔt rööp lueek cäŋ-ni libɛ-ni mat paan tap tiɛk̈ dɛjuac-ni puuk jɔam GEN PL kɔt-ni riöp-ni lueek-ni cäŋ-ni lipɛ-ni mät-ni paa-ni tääp-ni tiɛk̈-ni dɛy-ni puɔk-ni jiam-ni LOC PL kɔt-ni riöp-ni luaak cääŋ-ni lipɛ-ni mät-ni paa-ni tap-ni tiɛk̈-ni dɛy-ni puuk-ni jiam-ni ‘tamarind’ ‘fingernail’ ‘barn’ ‘sun’ ‘needle’ ‘narrows’ ‘desert’ ‘valley’ ‘necklace’ ‘plant’ ‘ash’ ‘wind’

A B B A B B A B B A B B A B B A B C A B C A B C A B C A B C A B C A B C C B D C C B C C D C D C C D E A A A A A C B B C A B B D D D D A E D E E

NOM SG tat päm mök dit dɛl ŋɔp jöŋ bɔɔr kɔryɔm thok gök buäw GEN SG taat-kä paam mok diɛt dɛɛ̈l̈ ŋɔäp jiöŋ-kä bɔr-kä kɔryɔam thuɔk gɔk-kä buɔ-kä LOC SG taat-kä paam mok diɛt dɛɛ̈l̈ ŋɔääp jioŋ bɔär-kä kɔryiöm-kä thɔk gok-kä bɔw-kä NOM PL tät pääm möök diit det ŋɔp-ni jöŋ-ni bɔr-i kɔryɔm-ni thuuk göök bɔɔw GEN PL taat-ni pääm-ni möök-ni diin dëët-ni ŋɔp-ni jöŋ-ni bɔr-i kɔryɔam-ni thuuk-ni gök-kä bɔäw-ni LOC PL taat-ni paam-ni mok-ni diit-ni deet ŋɔp-ni jioŋ-ni bɔär-i kɔryɔam-ni thuuk-ni gook-ni bɔäw-ni ‘buttock’ ‘mountain’ ‘buffalo’ ‘song’ ‘goat’ ‘kind of tree’ ‘table’ ‘water plant’ ‘locust’ ‘mouth’ ‘bag’ ‘kind of tree’

Documents

Paradigmatic opacity in Nuer - uni-leipzig.de · Matthew Baerman, Surrey Morphology Group, University of Surrey [email protected] revised post-conference version of 14/7/09 1